Large-scale geographical and historical factors can strongly affect species richness within ecological communities (local richness), because most communities are at least partially open to immigration from regional species pools. To completely understand local richness patterns, it is important to distinguish local explanations, such as spatiotemporal heterogeneity, from larger scale explanations for regional enrichment, and to integrate these explanations with traditional niche-based notions (e.g., resource partitioning, limiting similarity, limited membership, etc.). Here we use regression models to evaluate the sensitivity of local richness in reef-building corals to two local and six regional variables at multiple geographical scales. The local variables are depth and habitat, and the regional variables are the number of species and genera in a region, average age of genera, distance to the equator, distance to the nearest high-diversity region, and a dummy variable contrasting the Indo-Pacific and Atlantic provinces. At the largest geographical scale (Indo-Pacific and Atlantic provinces combined), we find that local richness varies significantly with depth, habitat, and five regional variables. The relative sensitivity of local richness to local and regional variables (61 and 39%, respectively) is similar to that at two smaller geographical scales: (1) the Indo-Pacific and (2) the speciose central Indo-Pacific. However, when only depauperate regions of the Indo-Pacific are considered, regional variables contribute more strongly to local richness, explaining 95% of the variation in this attribute. The analysis indicates that Indo-Pacific coral assemblages in both speciose and depauperate regions are unsaturated, as local richness consistently increases with the size of the regional species pool. Comparable regional effects are totally absent in the Atlantic province, where local richness is sensitive only to depth and habitat. This result is consistent with the assessment that regional species distributions are relatively homogeneous within this province. We interpret this lack of regional heterogeneity as a historical consequence of climatic and geologic events, which isolated this province from the Indo-Pacific in the Pliocene and caused major extinctions among predominantly stenotopic species 1-4 x 10(6) yr BP. In general, our results indicate that geographical scale is an important factor to consider in comparative regional studies of ecological communities. Furthermore, we suggest that purely local spatiotemporal heterogeneity is insufficient to explain the sensitivity of local richness to large-scale geographical and historical variables.